Ham et al. Ann. Intensive Care (2019) 9:63 https://doi.org/10.1186/s13613-019-0536-5

RESEARCH Open Access Sensitivity to II dose in patients with vasodilatory : a prespecifed analysis of the ATHOS‑3 trial Kealy R. Ham1*, David W. Boldt2, Michael T. McCurdy3, Laurence W. Busse4, Raphael Favory5,6, Michelle N. Gong7, Ashish K. Khanna8, Stefan N. Chock9, Feng Zeng10, Lakhmir S. Chawla10, George F. Tidmarsh10 and Marlies Ostermann11

Abstract Background: Early clinical data showed that some patients with vasodilatory shock are responsive to low doses of 1 1 angiotensin II. The objective of this analysis was to compare clinical outcomes in patients requiring 5 ng kg− min− 1 1 ≤ angiotensin II at 30 min ( 5 ng kg− min− subgroup) to maintain mean arterial pressure (MAP) 75 mmHg versus ≤ 1 1 1 1 ≥ patients receiving > 5 ng kg− min− angiotensin II at 30 min (> 5 ng kg− min− subgroup). Data from angiotensin II-treated patients enrolled in the ATHOS-3 trial were used. 1 1 Results: The subgroup of patients whose angiotensin II dose was down-titrated from 20 ng kg− min− at treat- 1 1 ment initiation to 5 ng kg− min− at 30 min (79/163) had signifcantly lower endogenous serum angiotensin II ≤ 1 1 levels and -equivalent doses and signifcantly higher MAP versus the > 5 ng kg− min− subgroup 1 1 (84/163). Patients in the 5 ng kg− min− subgroup were more likely to have a MAP response at 3 h versus those 1 1≤ in the > 5 ng kg− min− subgroup (90% vs. 51%, respectively; odds ratio, 8.46 [95% CI 3.63–19.7], P < 0.001). Day 28 1 1 1 1 survival was also higher in the 5 ng kg− min− subgroup versus the > 5 ng kg− min− subgroup (59% vs. 33%, respectively; hazard ratio, 0.48 ≤[95% CI 0.28–0.72], P 0.0007); multivariate analyses supported the survival beneft = 1 1 in patients with lower angiotensin II levels. The 5 ng kg− min− subgroup had a more favorable safety profle and ≤ 1 1 lower treatment discontinuation rate than the > 5 ng kg− min− subgroup. 1 1 Conclusions: This prespecifed analysis showed that down-titration to 5 ng kg− min− angiotensin II at 30 min is an early predictor of favorable clinical outcomes which may be related to≤ relative angiotensin II insufciency. Keywords: Shock, , Angiotensin II, Dose response,

Background beyond 28 days [4–7]. Additionally, treatment with Shock is characterized by inadequate organ , norepinephrine may cause immunosuppression and which, if not rapidly corrected, leads to multiorgan fail- seriously complicate the clinical course in patients with ure and death [1]. International guidelines recommend septic shock, the most common form of shock [8, 9]. as frst-line vasopressor agents [2, 3]. High-dose catecholamines are associated with worse Outcomes remain unacceptable because approximately outcomes and adverse events in patients who require half of patients with vasodilatory shock do not survive higher doses to achieve targeted hemodynamic param- eters [9, 10]. An alternative noncatecholamine agent, vasopressin [3], is efective in some patients and may *Correspondence: [email protected] allow the lowering of dose, but it has 1 Division of Critical Care , Regions Hospital, University of Minnesota, 640 Jackson Street, St. Paul, MN 55101, USA not been shown to improve survival [3]. Further, fewer Full list of author information is available at the end of the article than 50% of patients in shock generate a blood pres-

Feng Zeng: At the time of study/manuscript development sure response to treatment with vasopressin [11], and

© The Author(s) 2019. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat​iveco​mmons​.org/licen​ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Ham et al. Ann. Intensive Care (2019) 9:63 Page 2 of 10

vasopressin may cause cardiac toxicity, , Methods and mesenteric ischemia at higher doses [3, 12]. Study design Angiotensin II is a naturally occurring peptide with Complete details of the ATHOS-3 study (registered Jan- strong vasopressor activity [13]. A pilot study indicated uary 2015, https​://clini​caltr​ials.gov/ct2/show/NCT02​ that the addition of synthetic human angiotensin II to 33884​3; principal investigator: George Tidmarsh) have catecholamine and vasopressin therapy maintained been previously reported [15]. Briefy, ATHOS-3 was mean arterial pressure (MAP) in patients with vasodila- a placebo-controlled study that evaluated the safety tory shock, while allowing for a reduction in catecho- and efcacy of angiotensin II as an addition to back- lamine dosing [14]. In addition, the pilot study noted ground vasopressor therapy in patients with catechola- that 20% of treated patients were exquisitely sensitive mine-resistant vasodilatory shock. Randomization was to angiotensin II. Given the risk of higher-dose cat- stratifed according to MAP at screening (< 65 mmHg echolamines, the ability to approach or restore nor- vs. ≥ 65 mmHg) and Acute Physiology and Chronic mal hemodynamic function with low doses of systemic Health Evaluation II (APACHE II) score (≤ 30, 31 to 40, vasopressor(s) is an important clinical goal. vs. ≥ 41) on a scale of 0 to 71, with higher scores indicat- Following the pilot study, the randomized, double- ing greater disease severity. blind, phase 3 Angiotensin II for the Treatment of Tis registrational study was sponsored by La Jolla High-Output Shock (ATHOS-3) trial (ClinicalTrials. Pharmaceutical Company, was conducted under a special gov, NCT02338843) demonstrated that angiotensin II protocol agreement with the US Food and Drug Admin- signifcantly increased MAP and provided a catecho- istration, and was approved by the appropriate national lamine-sparing efect versus placebo in patients with authorities and local research ethics boards. Te study catecholamine-resistant vasodilatory shock [15]. In this was conducted in accordance with current Good Clini- study, patients receiving a stable dose of vasopressor cal Practice Guidelines, applicable local regulations, and were treated with a starting dose of 20 ng kg−1 min−1 the ethical principles described in the Declaration of angiotensin II that could be adjusted every 5 min to Helsinki. achieve a MAP of ≥ 75 mmHg by hour 3. By 30 min, the angiotensin II dose had been down-titrated from Patients −1 −1 20 to ≤ 5 ng kg min for 79 of 163 (48%) patients in ATHOS-3 enrolled adult patients (aged ≥ 18 years) the treatment arm; analyses comparing this population with vasodilatory shock, defned as central venous with those with a dose of > 5 ng kg−1 min−1 at 30 min oxygen saturation > 70% with central venous pres- were prespecifed. In a study by Fliser et al. [16], dos- sure > 8 mmHg or cardiac index > 2.3 L min−1 m2, with ing with 1.5 ng kg−1 min−1 of angiotensin II resulted catecholamine-resistant hypotension (patients receiv- in serum levels of ~ 50 pg/mL, just beyond the upper ing > 0.2 µg kg−1 min−1 of norepinephrine or norepineph- physiologic range for angiotensin II; therefore, admin- rine-equivalent dose for 6 to 48 h prior to enrollment to istration of 1.5 to 5 ng kg−1 min−1 angiotensin II is maintain a MAP of 55 to 70 mmHg) after adequate fuid expected to result in serum levels of angiotensin II that . approximate physiologic levels in a normotensive indi- vidual [16, 17]. Doses of > 5 ng kg−1 min−1 are expected Treatment to result in serum levels in the pharmacologic range Enrolled and consented patients were randomly assigned [16–18]. to angiotensin II or saline placebo. Te starting dose We hypothesized that patients most responsive to angi- of angiotensin II was 20 ng kg−1 min−1 and could be otensin II may have had a defect in angiotensin-convert- adjusted every 5 min to achieve a MAP of ≥ 75 mmHg by ing enzyme function resulting in diminished angiotensin hour 3. Te maximum permitted dose of angiotensin II II levels. We further hypothesized that these low angio- from hour 0 to 3 was 200 ng kg−1 min−1. Standard-of-care tensin II levels could be easily corrected to normal physi- vasopressors were held constant unless increases were ologic levels by direct exogenous supplementation with proscribed for safety reasons. From hours 3 to 48, dose of the low dose of angiotensin II. study drug could be adjusted to 1.25 to 40 ng kg−1 min−1 To further characterize the clinical beneft of angioten- to maintain a MAP of 65 to 70 mmHg, and standard- sin II and to explore these hypotheses, we compared vari- of-care vasopressors could be down-titrated. Tere was ous efcacy and safety parameters in patients receiving a protocol-mandated down-titration of angiotensin II angiotensin II in ATHOS-3 who were down-titrated to at hour 48. If, upon discontinuation, the subjects’ car- −1 −1 doses ≤ 5 ng kg min angiotensin II at 30 min versus diovascular (CV) Sequential Organ Failure Assessment patients who required > 5 ng kg−1 min−1 angiotensin II at (SOFA) score was 4, the investigator could restart treat- 30 min [15]. ment. During this treatment phase, the dose adjustment Ham et al. Ann. Intensive Care (2019) 9:63 Page 3 of 10

criteria remained the same as previously. Angiotensin more likely to have a higher baseline MAP (P < 0.0001) −1 −1 II was down-titrated gradually, by ≤ 10 ng kg min and a lower baseline norepinephrine-equivalent dose −1 −1 every 15 min, when being discontinued and standard-of- (P = 0.0049). Patients in the ≤ 5 ng kg min angio- care vasopressors were titrated according to institutional tensin II subgroup were also more likely to have lower protocols. Te protocol-specifed maximum duration of endogenous levels of angiotensin I (P = 0.0058) and angi- angiotensin use was 7 days. otensin II (P = 0.0009) at baseline (Table 1). Outcomes Efcacy in angiotensin II dose groups Efcacy outcomes included the proportion of patients MAP response at hour 3 who achieved a MAP response at hour 3, change in nor- A signifcantly higher proportion of patients receiv- −1 −1 epinephrine-equivalent dose from baseline, mortality ing ≤ 5 ng kg min angiotensin II at 30 min achieved to day 28, and safety. Te MAP response at hour 3 was a MAP response at hour 3 (≥ 75 mmHg or ≥ 10 mmHg defned as achieving ≥ 75 mmHg or an increase of ≥ 10 increase in MAP) compared with patients requir- mmHg from baseline at hour 3 without an increase in ing > 5 ng kg−1 min−1 angiotensin II at 30 min (89.9% standard-of-care vasopressors prior to hour 3. [95% CI 81.0–95.5%] vs. 51.2% [95% CI 40.0–62.3%], respectively; unadjusted odds ratio, 8.46 [95% CI 3.63– Statistical methods 19.74]; P < 0.001; Fig. 1a). Univariate analyses within sub- Te diference in outcomes based on angiotensin II dose groups of patients showed a consistent trend for higher at 30 min was a prespecifed analysis. No patients dis- MAP response in patients receiving 5 ng kg−1 min−1 continued treatment prior to 30 min, and all 163 patients angiotensin II at 30 min. Tis included subgroups defned receiving angiotensin II were included. by the quartiles of baseline angiotensin II (measured in Te association of angiotensin II dose with efcacy pg/mL). Statistically signifcant treatment benefts were parameters was evaluated using univariate and multivari- largest in patients with lower baseline angiotensin II lev- ate analyses to adjust for potential imbalances between els (Fig. 1b). Multivariate analyses of MAP response at the two dose groups. Time-to-event endpoints including hour 3, adjusting for baseline disease covariates, were all-cause survival were summarized using Kaplan–Meier also signifcant for angiotensin II dose at 30 min (odds estimates, and unadjusted group comparisons were ratio, 9.09; 95% CI 3.53–23.4; P ≤ 0.0001). conducted using the log-rank test. Multivariate time- to-event analyses were conducted using proportional Norepinephrine‑equivalent doses and SOFA score −1 −1 hazards modeling. Binary outcomes (e.g., hour 3 MAP Patients in the ≤ 5 ng kg min angiotensin II sub- responders) were analyzed by chi-square test for unad- group received lower norepinephrine-equivalent doses justed analyses and by logistic regression for multivariate from hours 3 to 48 (mean [SD] dose 0.16 ng kg−1 min−1 analyses. [0.242]) than patients in the > 5 ng kg−1 min−1 angioten- For multivariate analyses, the following dichotomized sin II subgroup (0.51 ng kg−1 min−1 [0.750]). At hour 48, −1 −1 baseline covariates were adjusted for age ≥ 65 years, 41 of 79 (52%) patients in the ≤ 5 ng kg min angioten- sex, MAP < 65 mmHg, APACHE II score > 30, albu- sin II subgroup had discontinued all other vasopressors min level < 2.5 g/dL, Model for End-Stage Liver Disease versus 25 of 84 (30%) patients in the > 5 ng kg−1 min−1 score ≥ 30, chest radiograph fnding of acute respiratory subgroup. Early sensitivity to angiotensin II was also distress syndrome (ARDS), and norepinephrine-equiva- associated with a signifcantly greater improvement in −1 −1 lent dose ≥ 0.5 μg kg min . Additional analyses were total SOFA score and CV SOFA score from screening to conducted, including baseline angiotensin I less than hour 48 (Table 2). median and angiotensin II less than median. Survival at day 28 Results Survival at day 28 was 67% (95% CI 56–76%) in patients Baseline demographic and disease characteristics −1 −1 down-titrated to ≤ 5 ng kg min angiotensin II at Of the 163 patients who were randomly assigned to 30 min compared with 41% (95% CI 31–52%) in patients receive angiotensin II, 79 patients (48.5%) were receiv- who received > 5 ng kg−1 min−1 angiotensin II at 30 min −1 −1 ing ≤ 5 ng kg min angiotensin II at 30 min, and 84 (relative risk, 0.45; 95% CI 0.28–0.72; P = 0.0007; patients (51.5%) were receiving > 5 ng kg−1 min−1 angio- Fig. 2). Univariate analyses within subgroups of patients tensin II at 30 min. Baseline demographic and disease showed a consistent trend for higher day 28 survival in −1 −1 characteristics of these 2 subgroups are shown in Table 1. patients receiving ≤ 5 ng kg min angiotensin II at In general, patients whose angiotensin II dose had been 30 min. For example, survival was signifcantly higher −1 −1 −1 −1 down-titrated at 30 min to ≤ 5 ng kg min were for patients receiving ≤ 5 ng kg min angiotensin Ham et al. Ann. Intensive Care (2019) 9:63 Page 4 of 10

Table 1 Baseline disease and demographic characteristics in analysis population

1 1 Angiotensin II > 5 ng kg− min− (n 84) Angiotensin = 1 1 II 5 ng kg− min− (n≤ 79) = Age, n (%) < 65 years 47 (56.0) 43 (54.4) 65 years 37 (44.0) 36 (45.6) ≥ Sex, n (%) Male 47 (56.0) 45 (57.0) Female 37 (44.0) 34 (43.0) Race, n (%) White 69 (82.1) 66 (83.5) Nonwhite 15 (17.9) 13 (16.5) Body mass index, n (%) n 83 n 78 = = < 18.5 kg/m2 3 (3.6) 4 (5.1) 18.5 to < 25 kg/m2 22 (26.5) 21 (26.9) ≥ 25 to < 30 kg/m2 20 (24.1) 22 (28.2) ≥ 30 kg/m2 38 (45.8) 31 (39.7) ≥ Geographic region, n (%) Australia/New Zealand 17 (20.2) 11 (13.9) Europe 10 (11.9) 9 (11.4) USA/Canada 57 (67.9) 59 (74.7) Albumin, n (%) n 78 n 76 = = < 2.5 g/dL 56 (71.8) 47 (61.8) 2.5 g/dL 22 (28.2) 29 (38.2) ≥ Cause of shock, n (%) 68 (81.0) 59 (74.7) Likely sepsis 8 (9.5) 12 (15.2) Vasoplegia 3 (3.6) 7 (8.9) Other 5 (6.0) 1 (1.3) Baseline MAP,* n (%) < 65 mmHg 37 (44.0) 15 (19.0) 65 mmHg 47 (56.0) 64 (81.0) ≥ Baseline APACHE II score, n (%) 30 54 (64.3) 51 (64.6) ≤ 31–40 29 (34.5) 21 (26.6) 41 1 (1.2) 7 (8.9) ≥ Norepinephrine-equivalent dose 6 h before randomization, mean, µg/kg/min 0.52 (0.301) 0.45 (0.377) (SD) Baseline norepinephrine-equivalent dose,* n (%) < 0.5 µg/kg/min 50 (59.5) 67 (84.8) 0.5 µg/kg/min 34 (40.5) 12 (15.2) ≥ Baseline angiotensin I,* n (%) n 74 n 71 = = Mean (SD) 678.9 (810.8) 618.0 (1332.5) Median (range) 346.0 (10.5–3730.0) 164 (10.5–9180.0) < 72.3 pg/mL 12 (16.2) 24 (33.8) 72.3 to < 253 pg/mL 14 (18.9) 19 (26.8) 253 to < 676 pg/mL 25 (33.8) 14 (19.7) 676 pg/mL 23 (31.1) 14 (19.7) ≥ Baseline angiotensin II,* n (%) n 74 n 70 = = Mean (SD) 420.8 (680.3) 128.3 (199.1) Median (range) 157.5 (10.5–3340.0) 61.3 (10.5–1090.0) Ham et al. Ann. Intensive Care (2019) 9:63 Page 5 of 10

Table 1 (continued)

1 1 Angiotensin II > 5 ng kg− min− (n 84) Angiotensin = 1 1 II 5 ng kg− min− (n≤ 79) = < 23.85 pg/mL 14 (18.9) 21 (30.0) 23.85 to < 83.75 pg/mL 11 (14.9) 21 (30.0) 83.75 to < 299.5 pg/mL 25 (33.8) 21 (30.0) 299.5 pg/mL 24 (32.4) 7 (10.0) ≥ Exposure to angiotensin-converting enzyme inhibitors 7 (8.3) 8 (10.1) Exposure to angiotensin II receptor blockers 9 (10.7) 2 (2.5) Radiographic fnding of ARDS 23 (27.4) 17 (21.5) Cardiac index (L/min/m2) n 36 n 33 = = Mean (SD) 3.5 (1.07) 3.1 (0.73) Median (range) 3.1 (2.3–6.4) 3.0 (2.1–5.4) ScvO (%) n 59 n 61 2 = = Mean (SD) 78.2 (8.14) 77 (9.64) Median (range) 79.0 (53–99) 75.9 (45–99) Central venous pressure (mmHg) n 64 n 62 = = Mean (SD) 14.3 (5.8) 13.2 (4.1) Median (range) 13.0 (5–35) 12.0 (6–29) SOFA score Mean (SD) 12.0 (2.8) 11.5 (2.8) Median (range) 12.0 (5–18) 12.0 (5–18)

ARDS, acute respiratory distress syndrome; MAP, mean arterial pressure; ­ScvO2, central venous oxygen saturation; SD, standard deviation; SOFA, Sequential Organ Failure Assessment All n’s are as in the table header, unless otherwise specifed Exposure to angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers within week prior to starting study *Statistically signifcant, P < 0.01

relatively low baseline angiotensin II levels (< 23.85 pg/ II regardless of baseline MAP (MAP ≤ 65 mmHg vs. MAP > 65 mmHg) (Table 3). In patient subsets defned mL) (Table 3). 1 1 by quartiles of baseline angiotensin I or II levels, a higher Patients receiving > 5 ng kg− min− angiotensin II at −1 −1 30 min were more likely to have serious adverse events, proportion of patients treated with ≤ 5 ng kg min angiotensin II were alive at day 28 than patients receiv- moderate to severe adverse events, or adverse events ing > 5 ng kg−1 min−1 angiotensin II. In multivariate resulting in discontinuation, compared with patients −1 −1 analyses to account for imbalances between dose groups, receiving ≤ 5 ng kg min angiotensin II at 30 min −1 −1 (Table 4). Te most commonly reported treatment- patients treated with ≤ 5 ng kg min angiotensin II were signifcantly more likely to survive than those emergent adverse events (reported in > 10% of patients receiving > 5 ng kg−1 min−1 (relative risk, 0.53; 95% CI in either dosing subgroup) included septic shock, atrial fbrillation, multiorgan failure, hypotension, thrombocy- 0.32–0.86; P = 0.011). topenia, and hypokalemia (Additional fle 1: Table S1). Treatment exposure and safety in angiotensin II dose groups Discussion In otherwise healthy patients, the plasma concentration Seventy-nine of 163 (48%) patients were down-titrated of angiotensin II is ~ 5 to 35 pg/mL [19]. In the criti- from an initial angiotensin II dose of 20 ng kg−1 min−1 cally ill patients with studied herein, −1 −1 to a dose of ≤ 5 ng kg min at 30 min following initia- the mean serum concentration of angiotensin II in tion of study drug. Tis subset of patients was more likely 1 1 the > 5 ng kg− min− angiotensin II subgroup was 420.8 to have a MAP response, increased 28-day survival, and −1 −1 (± 680.4) pg/mL at baseline and in the ≤ 5 ng kg min better safety profle when considering grade 3/4 and seri- angiotensin II subgroup was 128.3 (± 199.1) pg/mL at ous AEs. Tese data may provide health care practition- baseline. Survival was signifcantly higher for patients ers an early indication of which patients are most likely to −1 −1 in the ≤ 5 ng kg min angiotensin II subgroup with respond to angiotensin II. Ham et al. Ann. Intensive Care (2019) 9:63 Page 6 of 10

a 100

80 Odds ratio, 8.46 [95% CI, 3.63–19.7] P < 0.001 P Response, %

60

40

20 Proportion of Patients With MA 51.2 89.9 0 −1 −1 >5 ng kg−1 min−1 ≤5 ng kg min

Angiotensin II Dose

>5 ng kg−1 min−1 P = 0.0010 b ≤5 ng kg−1 min−1 P = 0.0097 P = 0.0531 % 100 100.0 90.5 85.7 P = 0.2331

75 71.4

57.1 54.5 56.0

50 45.8

25

n 14 21 11 21 25 21 24 7 0 Proportion of Patients With MAP Response, <23.85 23.85 to <83.75 83.75 to <299.5 ≥299.5

Baseline angiotensin II, pg/mL 1 1 Fig. 1 Proportion of patients who had a MAP response at hour 3 by their angiotensin II dose, at 30 min following study initiation; > 5 ng kg− min− 1 1 versus 5 ng kg− min− (a). Proportion of patients who had a MAP response at hour 3 by their angiotensin II dose at 30 min and quartiles of ≤ baseline angiotensin II (b). CI, confdence interval; MAP, mean arterial pressure Ham et al. Ann. Intensive Care (2019) 9:63 Page 7 of 10

Table 2 Change from screening to hour 48 in total SOFA hypothesis that patients sensitive to low levels of angio- score and CV SOFA score −1 −1 tensin II (≤ 5 ng kg min ) are more likely to have an Angiotensin Angiotensin P value angiotensin II insufciency. Nonetheless, supplemen- 1 1 1 1 II > 5 ng kg− min− II 5 ng kg− min− tation with exogenous angiotensin II in “angiotensin II (n 84) (n≤ 79) = = insufcient” patients helped rapidly restore MAP and Total SOFA score was associated with improved outcomes. Te possible Mean (SD) 2.6 (5.5) 0.6 (5.0) 0.002a reasons underlying sensitivity to exogenous angiotensin − 1 1 Median (range) 1.0 ( 6 to 15) 2.0 ( 10 to 15) II remain unclear, as patients in the 5 ng kg− min− − − − ≤ CV SOFA ­scoreb subgroup had lower baseline angiotensin II levels but Mean (SD) 1.33 (1.7) 2.2 (1.7) 0.004 also had higher baseline MAP and lower norepineph- − − Median (range) 0.0 ( 4 to 0) 3.0 (–4 to 0) rine-equivalent doses at baseline. In addition, no difer- − − −1 −1 CV, cardiovascular; MAP, mean arterial pressure; SOFA, Sequential Organ Failure ence existed between the > 5 ng kg min angiotensin −1 −1 Assessment; SD, standard deviation II subgroup and the ≤ 5 ng kg min angiotensin II a van Elteren Wilcoxon rank test of angiotensin II dose at 30 min stratifed by subgroup in characteristics potentially associated with randomization strata for MAP and APACHE II score impaired production/function of angiotensin II (such as b All patients had a CV SOFA score of 4 (highest risk) at screening based on inclusion/exclusion criteria. For patients missing a 48-hour assessment, prior use of angiotensin-converting enzyme inhibitors, the last observation was carried forward. In the event of death prior to the use of angiotensin II receptor blockers, or radiographic 48-hour assessment, the patient was assigned a CV SOFA score of 4. The van evidence of ARDS; Table 1). It is conceivable that angio- Elteren Wilcoxon rank test compared angiotensin II with placebo adjusting for randomization strata for MAP and APACHE II score tensin II sensitivity is linked to endothelial injury, which was not rigorously assessed in the current analyses or the ATHOS-3 study. Across the entire cohort, the deliv- −1 −1 Patients in the ≤ 5 ng kg min angiotensin II sub- ered dose of exogenous angiotensin II would overcome group had lower levels of endogenous baseline angio- any inherent angiotensin-converting enzyme defects, tensin II than their counterparts in the > 5 ng kg−1 min−1 so it is likely that attenuation of the MAP response to angiotensin II subgroup. However, the angiotensin II angiotensin II in the > 5 ng kg−1 min−1 subgroup may be levels for patients with shock are higher than those for infuenced by other factors, including down-regulation normal patients as previous studies have shown levels of of angiotensin I/II receptors in conditions such as sep- angiotensin II in normal adult plasma to range from 5 to sis and/or biofeedback to the production of vasodila- 35 pg/mL [19]. Te observations seen herein support our tory peptides [20–22]. Although our analysis does not

1.0 Angiotensin II dose at 30 mins: ≤5 ng kg −1 min−1 0.9 >5 ng kg −1 min−1 0.8

y 0.7 ++ 0.6

0.5 + 0.4 ++

Survival Probabilit 0.3

0.2 Dose Day 28 point estimate (95% CI) P value ≤5 ng kg−1 min−1 67.1% (55.6%–76.3%) 0.0007 0.1 >5 ng kg−1 min−1 41.4% (30.8%–51.7%)

0.0 012345687 91011121314151617118 92021222324 25 26 27 28 Time From Randomization, days Patients at risk, n: ≤5 ng kg−1 min−1 79 78 75 72 69 68 68 65 64 62 61 57 57 56 55 55 55 55 55 54 54 54 54 54 54 54 54 53 39 >5 ng kg−1 min−1 84 73 67 61 60 56 54 51 51 50 48 47 47 44 43 43 42 42 41 41 40 40 40 39 39 37 36 33 22 1 1 Fig. 2 Proportion of patients surviving at day 28 by their angiotensin II dose, at 30 min following study initiation; > 5 ng kg− min− 1 1 versus 5 ng kg− min− . CI, confdence interval ≤ Ham et al. Ann. Intensive Care (2019) 9:63 Page 8 of 10

Table 3 Univariate analysis of survival at day 28 for baseline covariates signifcantly diferent between the angiotensin II dose groups Patients, n, % (95% CI) Hazard ratio* (95% CI) P value Angiotensin Angiotensin 1 1 1 1 II > 5 ng kg− min− II 5 ng kg− min− ≤ Baseline MAP 65 mmHg 47, 44.7 (30.2–58.1) 64, 67.2 (54.2–77.2) 0.52 (0.29–0.92) 0.023 ≥ < 65 mmHg 37, 37.4 (22.1–52.7) 15, 66.7 (37.5–84.6) 0.38 (0.14–1.00) 0.041 Baseline norepinephrine-equivalent dose 1 1 < 0.5 µg kg− min− 50, 47.4 (33.0–60.5) 67, 68.7 (56.1–78.3) 0.53 (0.30–0.95) 0.030 1 1 0.5 µg kg− min− 34, 32.4 (17.6–48.0) 12, 58.3 (27.0–80.1) 0.45 (0.17–1.19) 0.100 ≥ Baseline angiotensin I < 72.3 pg/mL 12, 41.7 (15.2–66.5) 24, 66.7 (44.3–81.7) 0.49 (0.18–1.35) 0.160 72.3 to < 253 pg/mL 14, 42.9 (17.7–66.0) 19, 57.9 (33.2–76.3) 0.56 (0.21–1.48) 0.234 253 to < 676 pg/mL 25, 39.3 (20.5–57.6) 14, 85.7 (53.9–96.2) 0.18 (0.04–0.80) 0.011 676 pg/mL 23, 52.2 (30.5–70.0) 14, 50.0 (22.9–72.2) 1.02 (0.40–2.64) 0.962 ≥ Baseline angiotensin II < 23.85 pg/mL 14, 28.6 (8.8–52.4) 21, 71.4 (47.2–86.0) 0.29 (0.10–0.80) 0.011 23.85 to < 83.75 pg/mL 11, 36.4 (11.2–62.7) 21, 71.4 (47.2–86.0) 0.37 (0.12–1.09) 0.060 83.75 to < 299.5 pg/mL 25, 43.2 (23.5–61.5) 21, 57.1 (33.8–74.9) 0.64 (0.27–1.47) 0.285 299.5 pg/mL 24, 58.3 (36.4–75.0) 7, 57.1 (17.2–83.7) 1.01 (0.28–3.66) 0.993 ≥ CI, confdence interval; MAP, mean arterial pressure *Hazard ratio/log-rank test result of treatment efect within subgroup

Table 4 Summary of treatment-emergent adverse events and lower baseline norepinephrine-equivalent doses (regardless of causality) in either dosing subgroup than those who required > 5 ng kg−1 min−1. Accordingly, Angiotensin Angiotensin the benefcial efect seen in these patients may support 1 1 1 1 II > 5 ng kg− min− II 5 ng kg− min− the concept of using angiotensin II earlier in the course (n 84) (n≤ 79) = = of disease. Additionally, in prespecifed analyses, these Any grade, n (%) 76 (90.5) 66 (83.5) patients exhibited improved survival even after adjusting Grade 3/4, n (%) 64 (76.2) 43 (54.4) for baseline diferences in shock. Again, this would sug- Serious adverse events, 64 (76.2) 35 (44.3) gest that the use of angiotensin II at an early stage may n (%) be reasonable before a patient progresses to more severe Adverse events resulting 19 (22.6) 4 (5.1) refractory shock at very high norepinephrine-equivalent in discontinuation, n (%) doses. Fatal adverse events, 0 1 (1.3) Te study has several limitations. First, even though n (%) this was a prespecifed analysis, the subgroups were defned by an on-treatment variable that was likely to be related to drug efcacy and was associated with better prognostic characteristics, not all of which are measured account for all possible mechanisms that could account and controllable. However, no patients were discontin- for the diferences in MAP response between the two ued from treatment within 30 min, nor was there selec- −1 −1 subgroups, patients receiving ≤ 5 ng kg min angio- tion bias, as all patients receiving angiotensin II were tensin II at 30 min represent a subset of patients who may included in the analyses. Te on-treatment variable of be more likely to safely recover from shock and who can angiotensin II dose was adjusted based on MAP respon- be identifed soon after treatment initiation (at 30 min). siveness, so higher hour 3 response rates in patients who Tis observation may allow clinicians to rapidly identify ultimately required lower hour 3 doses of angiotensin those patients who may beneft most from the continued II were expected. However, dose adjustment was made use of angiotensin II. based on prespecifed, objective, and easily measurable −1 −1 Tose patients who required ≤ 5 ng kg min at criteria (i.e., MAP), adding validity to the study fndings. 30 min had less severe shock (higher baseline MAP) Te observed improvement in day 28 survival in patients Ham et al. Ann. Intensive Care (2019) 9:63 Page 9 of 10

Funding −1 −1 receiving ≤ 5 ng kg min of angiotensin II is not These analyses and the original study were funded by La Jolla Pharmaceutical directly linked to angiotensin II dose, but it is expected Company. that patients with better prognostic characteristics may Availability of data and material be more likely to demonstrate immediate responsive- The datasets used and analyzed during the current analyses are available from ness. As a result, there were signifcant imbalances the corresponding author on reasonable request. in key disease and previous treatment characteristics Ethics approval and consent to participate between the two groups that confound the interpreta- The ATHOS-3 study was approved by the appropriate national authorities and tion of these analyses. For example, patients receiv- local research ethics boards. The study was conducted in accordance with the ing 5 ng kg−1 min−1 angiotensin II had signifcantly current Good Clinical Practice Guidelines, applicable local regulations, and ≤ ethical principles described in the Declaration of Helsinki. lower baseline norepinephrine-equivalent dose and sig- nifcantly higher baseline MAP compared with patients Consent for publication in the subgroup receiving > 5 ng kg−1 min−1. To account Not applicable. for the baseline diferences, we performed regression Competing interests analyses adjusting for the baseline characteristics. How- KRH reports no confict of interest. DWB received consulting payment from La ever, there may be unmeasured characteristics that were Jolla Pharmaceutical Company and received grant/payment to his institu- tion from La Jolla Pharmaceutical Company. MTM is on the speakers bureau unaccounted for. for La Jolla Pharmaceutical Company, and his institution received a grant for its role as a participating site in the ATHOS-3 trial. LWB reports payment for consulting work from La Jolla Pharmaceutical Company. RF reports nothing to Conclusions disclose. MNG reports payment to her institution for the conduct of the trials in the form of a contract to perform the study. AKK reports payment from La Despite these limitations, this prespecifed analysis of Jolla Pharmaceutical Company for consulting work and participation in the ATHOS-3 showed that, for a signifcant proportion of company’s speakers bureau. SNC reports nothing to disclose. LSC and GFT are patients with vasodilatory shock, a low dose of angio- employees of La Jolla Pharmaceutical Company. MO and FZ report nothing to disclose. tensin II may be efective and is safe, and may provide an early indication of which patients are more likely to Author details 1 respond to angiotensin II. Larger, prospective studies are Division of Critical Care Medicine, Regions Hospital, University of Minnesota, 640 Jackson Street, St. Paul, MN 55101, USA. 2 Division of Critical Care, Depart- needed to understand the reasons underlying the phe- ment of , University of California, Los Angeles, Los Angeles, CA, nomenon of hyper-responsiveness to low doses of angio- USA. 3 Department of Medicine, University of Maryland School of Medicine, 4 tensin II and to identify all responders. Such information Baltimore, MD, USA. Department of Medicine, Emory University, Atlanta, GA, USA. 5 Critical Care Center, CHU Lille, Lille, France. 6 Lille Infammation Research would be informative for clinical decision-making as well Center, University of Lille School of Medicine, Lille, France. 7 Department as help improve emergency management of patients with of Medicine, Montefore Healthcare Center, Albert Einstein College of Medi- 8 vasodilatory shock. cine, Bronx, NY, USA. Department of Anesthesiology, Section on Critical Care Medicine, Wake Forest School of Medicine, Winston-Salem, NC & Outcomes Research Consortium, Cleveland, OH, USA. 9 Department of , Sunrise Hospital, Las Vegas, NV, USA. 10 La Jolla Pharmaceutical Company, San Diego, CA, USA. 11 Department of Critical Care, Guy’s and St. Thomas Hospital, King’s Additional fle College London, London, UK. Received: 5 April 2019 Accepted: 20 May 2019 Additional fle 1: Table S1. Summary of Treatment-Emergent Adverse Events (regardless of causality) Reported in >5% of Patients in Either Dos- ing Subgroup.

References Abbreviations 1. Vincent JL, De Backer D. Circulatory shock. N Engl J Med. APACHE II: Acute Physiology and Chronic Health Evaluation II; ARDS: acute 2013;369:1726–34. respiratory distress syndrome; ATHOS-3: Angiotensin II for the Treatment of 2. Avni T, Lador A, Lev S, Leibovici L, Paul M, Grossman A. Vasopressors for High-Output Shock; CI: confdence interval; MAP: mean arterial pressure; SD: the treatment of septic shock: systematic review and meta-analysis. PLoS standard deviation; SOFA: Sequential Organ Failure Assessment. ONE. 2015;10:e0129305. 3. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, Opal SM, Acknowledgements Sevransky JE, Sprung CL, Douglas IS, Jaeschke R, Osborn TM, Nunnally Medical writing and editorial assistance was provided by Jerome Sah, PhD, of ME, Townsend SR, Reinhart K, Kleinpell RM, Angus DC, Deutschman ApotheCom, Yardley, PA, USA. This assistance was funded by La Jolla Pharma- CS, Machado FR, Rubenfeld GD, Webb S, Beale RJ, Vincent JL, Moreno ceutical Company, San Diego, CA, USA. R. Surviving Sepsis Campaign: international guidelines for manage- ment of severe sepsis and septic shock, 2012. Intensive Care Med. Authors’ contributions 2013;39:165–228. KRH, DWB, MTM, and LWB helped interpret the results of the analyses and 4. De Backer D, Biston P, Devriendt J, Madl C, Chochrad D, Aldecoa C, Bras- were major contributors in writing of the manuscript. RF, MNG, AKK, SNC, and seur A, Defrance P, Gottignies P, Vincent JL, Investigators SI. Comparison MO were major contributors in providing interpretation of the data. FZ, LSC, of dopamine and norepinephrine in the treatment of shock. N Engl J and GFT were involved in the acquisition and analysis of the data. All authors Med. 2010;362:779–89. read and approved the fnal manuscript and have agreed to be accountable 5. Andreis DT, Singer M. Catecholamines for infammatory shock: a Jekyll- for the work. All authors read and approved the fnal manuscript. and-Hyde conundrum. Intensive Care Med. 2016;42:1387–97. Ham et al. Ann. Intensive Care (2019) 9:63 Page 10 of 10

6. Hessler M, Arnemann PH, Seidel L, Kampmeier T, Rehberg S, Ertmer C. investigators: angiotensin II for the treatment of vasodilatory shock. N ATHOS-3 and the knights of the round table–the search for the holy grail Engl J Med. 2017;377:419–30. of vasopressors. J Thorac Dis. 2017;9:2788–91. 16. Fliser D, Arnold U, Kohl B, Hartung R, Ritz E. Angiotensin II enhances 7. Schmittinger CA, Torgersen C, Luckner G, Schroder DC, Lorenz I, Dunser insulin sensitivity in healthy volunteers under euglycemic conditions. J MW. Adverse cardiac events during catecholamine vasopressor therapy: a Hypertens. 1993;11:983–8. prospective observational study. Intensive Care Med. 2012;38:950–8. 17. Wolf RL, Mendlowitz M, Gitlow SE, Naftchi N. Metabolism of angiotensin 8. Stolk RF, van der Poll T, Angus DC, van der Hoeven JG, Pickkers P, Kox II-I-131 in normotensive and hypertensive human subjects. Circulation. M. Potentially inadvertent immunomodulation: norepinephrine use in 1961;23:754–8. sepsis. Am J Respir Crit Care Med. 2016;194:550–8. 18. Fliser D, Dikow R, Demukaj S, Ritz E. Opposing efects of angiotensin II 9. Yamamura H, Kawazoe Y, Miyamoto K, Yamamoto T, Ohta Y, Morimoto on muscle and renal blood fow under euglycemic conditions. J Am Soc T. Efect of norepinephrine dosage on mortality in patients with septic Nephrol. 2000;11:2001–6. shock. J Intensive Care. 2018;6:12. 19. Dusterdieck G, McElwee G. Estimation of angiotensin II concentration in 10. Martin C, Medam S, Antonini F, Alingrin J, Haddam M, Hammad E, Meys- human plasma by radioimmunoassay. Some applications to physiological signac B, Vigne C, Zieleskiewicz L, Leone M. Norepinephrine: not too and clinical states. Eur J Clin Invest. 1971;2:32–8. much, too long. Shock. 2015;44:305–9. 20. Bucher M, Hobbhahn J, Kurtz A. Nitric oxide-dependent down-regulation 11. Sacha GL, Lam SW, Duggal A, Torbic H, Bass SN, Welch SC, Butler RS, Bauer of angiotensin II type 2 receptors during experimental sepsis. Crit Care SR. Predictors of response to fxed-dose vasopressin in adult patients with Med. 2001;29:1750–5. septic shock. Ann Intensive Care. 2018;8:35. 21. Bucher M, Ittner KP, Hobbhahn J, Taeger K, Kurtz A. Downregula- 12. Bassi E, Park M, Azevedo LC. Therapeutic strategies for high-dose tion of angiotensin II type 1 receptors during sepsis. Hypertension. vasopressor-dependent shock. Crit Care Res Pract. 2013;2013:654708. 2001;38:177–82. 13. Correa TD, Takala J, Jakob SM. Angiotensin II in septic shock. Crit Care. 22. Chawla LS, Chen S, Bellomo R, Tidmarsh GF. Angiotensin converting 2015;19:98. enzyme defects in shock: implications for future therapy. Crit Care. 14. Chawla LS, Busse L, Brasha-Mitchell E, Davison D, Honiq J, Alotaibi Z, 2018;22(1):274. Senef MG. Intravenous angiotensin II for the treatment of high-output shock (ATHOS trial): a pilot study. Crit Care. 2014;18:534. 15. Khanna A, English SW, Wang XS, Ham K, Tumlin J, Szerlip H, Busse LW, Publisher’s Note Altaweel L, Albertson TE, Mackey C, McCurdy MT, Boldt DW, Chock Springer Nature remains neutral with regard to jurisdictional claims in pub- S, Young PJ, Krell K, Wunderink RG, Ostermann M, Murugan R, Gong lished maps and institutional afliations. MN, Panwar R, Hastbacka J, Favory R, Venkatesh B, Thompson BT, Bel- lomo R, Jensen J, Kroll S, Chawla LS, Tidmarsh GF, Deane AM. ATHOS-3